I am pretty sure C++ is absolutely worst choice for crypto. Python is probably the best (Electrum?).
Just so: “PyBC Generic Blockchain Library (and coin!)”
https://bitcointalksearch.org/topic/annlibraryv040-pybc-generic-blockchain-library-and-coin-252152And, to lay down a nominal scale for one dimension: the main characterisation in this XKCDforum post
http://forums.xkcd.com/viewtopic.php?f=11&t=108685 seems broadly accurate in terms of speed of general-purpose processing:
Tier 1 (fast-fast): C++, C, Ada, Fortran, ATS.
Tier 2 (fast): Java, LuaJIT, Julia, Haskell, Scala, Ocaml, C#, Go, Common Lisp (SBCL), Rust, Pascal, F#.
"Fringe" (fast-ish): Clojure, Racket, JavaScript (V8 and others), Dart, PyPy.
Tier 3 (intermediate): Erlang HiPE.
Tier 4 (slow): Erlang, PHP, Smalltalk (VisualWorks), Lua, Perl, CPython, Hack, Ruby.
Tier 5 (very slow): R, Matlab, Octave.
Some interesting angles in KnightExemplar's insightful response of an alternative and very plausible framing of Tier 0 and Tier 1 as specialised and hardware languages respectively:
http://forums.xkcd.com/viewtopic.php?f=11&t=108685#p3575512 - but unlikely to be news to those of a mining persuasion.
Scaling a p2p architecture to support serious mass adoption is going to be fun.
Cheers
Graham